• Publications
  • Influence
On Generating Characteristic-rich Question Sets for QA Evaluation
TLDR
This work is the first to generate questions with explicitly specified characteristics for QA evaluation, and it is shown that datasets constructed in this way enable finegrained analyses of QA systems.
DialSQL: Dialogue Based Structured Query Generation
TLDR
DialSQL is a dialogue-based structured query generation framework that leverages human intelligence to boost the performance of existing algorithms via user interaction and is capable of identifying potential errors in a generated SQL query and asking users for validation via simple multi-choice questions.
Improving Semantic Parsing via Answer Type Inference
TLDR
The possibility of inferring the answer type before solving a factoid question and leveraging the type information to improve semantic parsing is shown and it is observed that if the authors convert a question into a statement form, the LSTM model achieves better accuracy.
Learning to Navigate the Web
TLDR
DQN, deep reinforcement learning agent, with Q-value function approximated with a novel QWeb neural network architecture is trained with the ability of the agent to generalize to new instructions on World of Bits benchmark, on forms with up to 100 elements, supporting 14 million possible instructions.
Accurate Supervised and Semi-Supervised Machine Reading for Long Documents
TLDR
This work introduces a hierarchical architecture for machine reading capable of extracting precise information from long documents and evaluates models that can reuse autoencoder states and outputs without fine-tuning their weights, allowing for more efficient training and inference.
User Modeling for Task Oriented Dialogues
TLDR
This work designs a hierarchical sequence-to-sequence model that first encodes the initial user goal and system turns into fixed length representations using Recurrent Neural Networks (RNN), and develops several variants by utilizing a latent variable model to inject random variations into user responses to promote diversity in simulated user responses.
What It Takes to Achieve 100% Condition Accuracy on WikiSQL
TLDR
This paper focuses on the WHERE clause in SQL and proposes a solution that can reach up to 88.6% condition accuracy on the WikiSQL dataset.
Global Relation Embedding for Relation Extraction
TLDR
It is shown that the learned textual relation embedding can be used to augment existing relation extraction models and significantly improve their performance, most remarkably, for the top 1,000 relational facts discovered by the best existing model.
Recovering Question Answering Errors via Query Revision
TLDR
This work proposes to crosscheck the corresponding KB relations behind the predicted answers and identifies potential inconsistencies to improve the F1 score of STAGG, one of the leading QA systems, from 52.5% to 53.9% on WEBQUESTIONS data.
...
1
2
...